The Bitcoin Dilemma: Can the Internet be shutdown?

[JusticeForYou]

I think the real interesting question is what would happen in the case of a netsplit (used to be common in IRC) -- today one would probably be caused by a totalitarian regime implementing a countrywide firewall (google "great firewall of china").  First question is how well bitcoin could route around it?  

The second question comes from the observation that clearly after a few weeks passed we'd then have 2 independent perfectly consistent blockchains.  After a few weeks, months, or years, when the great firewall finally goes down, how does the bitcoin network rejoin?  In theory it would be possible to rejoin perfectly.  In practice, I think that anyone with a communications pathway (possibly even paper) could double-spend their coins on either side of the firewall, causing major chaos.

Very Good, This is testable and it doesn't need to be a country wide isolation.

Have 3 Bitcoin clients and 2 miner bitcoind exposed to the internet. Then disconnect the internet but continue transactions between the clients and allow the miners to continue to find blocks. The difficulty will drop and you will quickly start finding blocks. After 2 months, expose the network back to the internet and see what happens.

[DeathAndTaxes]

I think the real interesting question is what would happen in the case of a netsplit (used to be common in IRC) -- today one would probably be caused by a totalitarian regime implementing a countrywide firewall (google "great firewall of china").  First question is how well bitcoin could route around it? 

The second question comes from the observation that clearly after a few weeks passed we'd then have 2 independent perfectly consistent blockchains.  After a few weeks, months, or years, when the great firewall finally goes down, how does the bitcoin network rejoin?  In theory it would be possible to rejoin perfectly.  In practice, I think that anyone with a communications pathway (possibly even paper) could double-spend their coins on either side of the firewall, causing major chaos.

The smaller segment (i.e. china vs rest of world) will have a shorter blockchain.  When the network rejoins all the blocks from the smaller subnet will be oprhaned as nodes replace them with blocks from the larger subnet.

Any non double spend tx would simply revert to 0-confirm and would be included in the next block (or blocks because blocks will be very full depends on how long the split lasted).

All coinbase tx will be orphaned.  There is a 120 block limit on spending but a split of >120 blocks would mean any subsequent tx will be invalid. 

Intentional double spends would be the largest issue.

Smart thing to do in a network split like that would be to make a paper backup of your wallet and wait to rejoin the network.

[JusticeForYou]

I think the real interesting question is what would happen in the case of a netsplit (used to be common in IRC) -- today one would probably be caused by a totalitarian regime implementing a countrywide firewall (google "great firewall of china").  First question is how well bitcoin could route around it? 

The second question comes from the observation that clearly after a few weeks passed we'd then have 2 independent perfectly consistent blockchains.  After a few weeks, months, or years, when the great firewall finally goes down, how does the bitcoin network rejoin?  In theory it would be possible to rejoin perfectly.  In practice, I think that anyone with a communications pathway (possibly even paper) could double-spend their coins on either side of the firewall, causing major chaos.

The smaller segment (i.e. china vs rest of world) will have a shorter blockchain.  When the network rejoins all the blocks from the smaller subnet will be oprhaned as nodes replace them with blocks from the larger subnet.

Any non double spend tx would simply revert to 0-confirm and would be included in the next block (or blocks because blocks will be very full depends on how long the split lasted).

All coinbase tx will be orphaned.  There is a 120 block limit on spending but a split of >120 blocks would mean any subsequent tx will be invalid. 

Intentional double spends would be the largest issue.

Smart thing to do in a network split like that would be to make a paper backup of your wallet and wait to rejoin the network.

The man knows what he is talking about.

Now, can you foresee a Financial War (literally), i.e China vs. U.S.

[thezerg]

Its an interesting question.  The firewalled subset would not necessarily be the smaller blockchain if it used bitcoins heavily would it?  What I'm trying to say is that the side that everybody thinks should "win" (due to other criteria like being the "majority" of the world) might not actually be the bitcoin "winner".

Also, what about a malicious attempt to increase blockchain length?  I guess "they" could only double spend known coins so a malicious country would have to prepare by remembering every coin that passes through its system...

I need to read the papers and code I guess...

@DeathAndTaxes:  Its good to know that non double spent are separable from suspect coins...

what do you mean by: All coinbase tx will be orphaned.  There is a 120 block limit on spending but a split of >120 blocks would mean any subsequent tx will be invalid.

thx!

[MoonShadow]

I think the real interesting question is what would happen in the case of a netsplit (used to be common in IRC) -- today one would probably be caused by a totalitarian regime implementing a countrywide firewall (google "great firewall of china").  First question is how well bitcoin could route around it? 

This kind of thing has been discussed to great length in prior years. The end conclusion is that it is assumed that a network split of this magnatude cannot be maintained for any significant period of time.  Partly because of the nature of the bitcoin network, it only takes one live link between these two networks (between two bitcoin clients) to prevent the lesser side from dropping behind, and even that can be intermittent.  There is also the fact that there would be some rather creative people on both sides of the split with vested interests in undermining any intentional split.  A single thumbdrive with a copy of the longest blockchain smuggled into China would completely destroy the shorter blockchain and update the entire minority network in under an hour.

However, to our knowledge this kind of thing has never happened on any significant scale, so it's all speculation until it happens.  This is why I said that Iran might be a neat test case.

[MoonShadow]

Its an interesting question.  The firewalled subset would not necessarily be the smaller blockchain if it used bitcoins heavily would it?  What I'm trying to say is that the side that everybody thinks should "win" (due to other criteria like being the "majority" of the world) might not actually be the bitcoin "winner".

No.

Also, what about a malicious attempt to increase blockchain length? 

That's mathmaticly impossible.

I guess "they" could only double spend known coins so a malicious country would have to prepare by remembering every coin that passes through its system...

Any existing coins that could be spent on either side of the split would be limited in scope.  That said, there are methods for detecting a split.  Most importantly, there are methods for automaticly detecting that 1) there was a network split and 2) you're on the minority side.  Which is important because any double spend attacks are really attacks against the vendors on the minority side, because it's very very unlikely that a reconnection of the blockchain leaves the vendors on the majority side on the short end of the stick.

I need to read the papers and code I guess...

Or the whitepaper.  Bitcoin doesn't work like you are imagining that it does.

[MoonShadow]

what do you mean by: All coinbase tx will be orphaned.

Any blocks produced by miners caught on the minority side of the split will be over-written by the blocks produced on the majority side upon reconnection of the network split.  Thus, transactions included in those abandoned blocks will return to 0/unconfirmed and have to be reincluded into a new block.  Any 'coinbase' transactions (i.e. the transactions that gives the miner of a block 50 btc for his efforts) and any transactions based upon those same coinbase transactions that occur on the minority side of a split will be invalid after reconnection.  This is why coinbase transactions take 120 blocks to mature, while regular transactions take only 6; to limit the possibility that a temporary network split (due to non-malicious failures of the Internet) could be used in a follow up transaction that would prove invalid later.  The truth is, temporary network splits occur regularly, likely about once a day, but are expected and the bitcoin protocol has a method for self-correction.  If you mine solo, and ever get a block that the network rejects, it is because someone else also found a block at about the same time and beat you to half of the network.  However, if there is another tie on the two sides of the network, the split can continue until a block is found that isn't a tie, and then the side of the split that lost that race updates to the winning blockchain.  It's not uncommon for there to be two or three blocks in a row that 'tied' before a tie-breaker block was found; but I don't think that it's ever even hit 6 blocks in a row, much less 120.

[DeathAndTaxes]

The firewalled subset would not necessarily be the smaller blockchain if it used bitcoins heavily would it?  What I'm trying to say is that the side that everybody thinks should "win" (due to other criteria like being the "majority" of the world) might not actually be the bitcoin "winner".

All that matter is hashing power.  It is a mathematical certainty that given enough time the subnet with more hashing power will build the longer chain.  My example only refered to larger and smaller subnets.  The larger one would biuild a longer chain and thus overwrite the consensus of the smaller one.

Usage, number of tx, number of nodes, population, geography are all irrelivent.  When there is a split both sides of the split form a subnet and the one with larger hashing power will prevail when the split is resolved.

Also, what about a malicious attempt to increase blockchain length? 

There is no way to find blocks faster than brute force.  In a dispute the side with more hashing power will always win.

I guess "they" could only double spend known coins so a malicious country would have to prepare by remembering every coin that passes through its system...

I have no idea what you are saying here.

what do you mean by: All coinbase tx will be orphaned.  There is a 120 block limit on spending but a split of >120 blocks would mean any subsequent tx will be invalid.

1) All coinbase rewards on the smaller shorter chain will be orphaned.  So any coins earned directly by miners on the smaller chain will be erased. 

2) The Bitcoin network tries to limit the effect of "accidental double spends" due to orphaned chains by making generation tx unspendable for 120 blocks.  If the split is longer than 120 blocks then they could be spent.  Regardless of who is holding those coins when the network re-orgs that generation becomes invalid and thus any "downstream" tx are invalidated too.